Page 6 of 9
Posted: Wed Aug 09, 2006 12:32 am
I just viewed calibration movie 2062521V1, which has something that appears to come into focus below the surface (at the very top of the screen, in the middle). I clicked on it, and it was scored as incorrect. I don't think it was a track, but it looks like some of the features of interest that are appearing on the real movies. Don't mind my score going down (at this point one or two mistakes don't hurt me much) but I don't think it's a good idea to discourage people from clicking on tracks like this one--at least not until you know more about what you are really looking for.
Thanks for this great opportunity to help with your research.
Posted: Wed Aug 09, 2006 6:58 am
I'm clicking on movies with obvious dust particles and getting graded as incorrectly IDing the movie. This has happened at least five time this morning and it is the first time this has happened to me.
Posted: Wed Aug 09, 2006 7:03 am
I just found out what I was doing wrong. For some reason, on this morning only, when a particle appeared I would just click anywhere in the movie. I just did an experiment and found that it would only score a correct if the exact particle location was clicked.
Posted: Wed Aug 09, 2006 12:18 pm
Positive Calibration Images Easilly Recognisable
Posted: Wed Aug 09, 2006 1:51 pm
I can distinguish most of positive calibration images by recognisable posterization in soft gradients. They were either compressed to JPEG with lower bpp ratio or they were compressed twice or they were gamma-tuned to match the real samples.
IMHO it's bad because it's easy to train the brain to subconciously distinguish them and pay more attention for these calibration images than for the real samples.
Posted: Wed Aug 09, 2006 6:36 pm
JT while I agree that most of the callibration movies are easy to spot, many of the real movies share the same features. Just don't let yourself get lazy! I'm still hoping they'll recycle some of the first batch movies into callibration movies. And catch people snoozing.
Movie contains an image from a different gel plate
Posted: Thu Aug 10, 2006 12:34 am
I found a movie (1620542V1) that has images of two different gel plates. The very bottom picture in the focus movie is clearly different from the other pictures at the other focus levels. I saved printscreens from the two pictures if that helps.
Re: Movie contains an image from a different gel plate
Posted: Thu Aug 10, 2006 1:21 am
jdriscol wrote:I found a movie (1620542V1) that has images of two different gel plates. The very bottom picture in the focus movie is clearly different from the other pictures at the other focus levels. I saved printscreens from the two pictures if that helps.
There is more information on this subject in the FAQ forum.
Here is the link-> http://stardustathome.ssl.berkeley.edu/ ... .php?t=469
Specificity Score - bug or feature?
Posted: Thu Aug 10, 2006 8:41 am
Last night my specificity score was 99%. This morning when I logged in it was reset to be the exact same number as my sensitivity score 93.58%. Yesterday it was xx%. I appreciate the extra info provided by the xx.xx%, but what's with the specificity score? As I dusted today, the specificity is keeping pace with sensitivity.
Posted: Thu Aug 10, 2006 11:34 am
This morning my sensitivity and specificity measures were no longer correct.
Posted: Thu Aug 10, 2006 11:49 am
Had 100% specificity, 98% sensitivity (5 out of 406 missed) on 8/9. Sometime after that (last night), it changed to 97.74%/97.74%, with same number missed (so, the missed were still "sensitivity" type).
Just another data point for trouble shooting.
Thanks again to the team.
Re: Specificity Score - bug or feature?
Posted: Thu Aug 10, 2006 5:21 pm
CrazyCelt wrote:Last night my specificity score was 99%. This morning when I logged in it was reset to be the exact same number as my sensitivity score 93.58%. Yesterday it was xx%. I appreciate the extra info provided by the xx.xx%, but what's with the specificity score? As I dusted today, the specificity is keeping pace with sensitivity.
I see you fixed it. Thanks.
Posted: Fri Aug 11, 2006 12:26 pm
Encountered Issue #3 with movie 7699943V1. The Movie ID did not change even when the second image appeared. I did not know to mark it "bad focus" at the time but after reading the instructions to do so in this thread, I will do so next time it happens.
Posted: Wed Aug 16, 2006 3:29 pm
On Wednesday August 16 at approx. 6:15 pm est, my ranking was reset to zero. I (gailod) was # 12 on the list, with a score of 5723.
I also noticed that the number 1 ranked person is now gone from the list - they had a score of 10,001 - the first person to break 10,000 - now gone!
Hope this can be fixed soon.
Finding 'cheaters' or NOT!!!
Posted: Wed Aug 16, 2006 6:52 pm
On another thread this evening I found a note about finding so-called 'cheaters', which is apparently why I was removed from the top 15 rankings after much hard work. Here is my reply to Buttorworth:
A"fter viewing several thousand movies, I became frustrated with poor response time from the pc for loading the next movie. I would wait and wait. While waiting, and having plenty of time to look, I noticed that there are splash circlets, of very light , VERY light shades of grey. Concentric circles, irregular. I am observant and have a good eye. Like a stone throne into water, particles are leaving ripples as shades of grey in what we se online. Would we could see color! Very subtle ripples ion grey, but there. I discovered that upon the instant of seeing the initial phase of a movie, I could evaluate presence of splash/ripples and thus a PARTICLE, and reject it or explore it. My specificity is 99.93%, and the other about 95%. I do not waste time on waiting for the next movie to fully load when I do not see the tell-tale ripple. I click No-Track and move on. I am too quick and I do press no-track if I am not fully engaged. That happens. I accept it and move on, although I slap my head and yell 'D'oh!'!.
By profession I am a tester. I am trained to look for in/consistencies and discrepancies in results. I am very good at what I do. This is not cheating, but something you did not expect. Someone who can instantly spot potential for a track, and who knows immediately when there is no potential - or when there IS.